# Game Modes

MindDuel ships with three live modes and two more on the roadmap. Every mode uses the same trivia gate, escrow, and settlement logic — only the board behavior changes.

## Mode comparison

| Mode               | Status  | Board             | Twist                                                 | Turn timeout |
| ------------------ | ------- | ----------------- | ----------------------------------------------------- | ------------ |
| **Classic Duel**   | Live    | 3x3               | Pure baseline. Win with 3 in a row.                   | 24 hours     |
| **Shifting Board** | Live    | 3x3               | Entire board rotates every 3 rounds via slot entropy. | 24 hours     |
| **Scale Up**       | Live    | 3x3 -> 4x4 -> 5x5 | Board grows as correct answers accumulate.            | 24 hours     |
| **Blitz**          | Phase 2 | 3x3               | 5-minute on-chain timer per turn.                     | 5 minutes    |
| **Ultimate TTT**   | Stretch | 9x9 meta-grid     | Each cell of the meta-grid is itself a TTT board.     | TBD          |

A practice mode (`vs-ai`) also exists for free, no-stake play. AI logic runs entirely in the frontend; results are stored as practice matches and excluded from the main leaderboard.

## Classic Duel

The cleanest version of the game.

* 3x3 board, standard alignment rules: 3 in a row horizontally, vertically, or diagonally.
* If all 9 cells fill with no winner, the pot splits 50/50 (after fee).
* Inactive turn after 24 hours -> any signer can call `timeout_turn` to flip the turn.

Best mode for first-time players.

## Shifting Board

The board itself becomes a threat.

* Starts as a 3x3 game.
* Every 3 rounds, the entire board rotates one of four directions:

| `slot % 4` | Direction                             |
| ---------- | ------------------------------------- |
| 0          | Rows shift down (top wraps to bottom) |
| 1          | Rows shift up                         |
| 2          | Columns shift right                   |
| 3          | Columns shift left                    |

* The shift direction is `Clock::get()?.slot % 4` — deterministic, observable on-chain, no oracle needed.
* Pieces keep their owners after a shift. A near-win can suddenly disappear; a scattered opponent's pieces can suddenly form a line.

You have to play one move ahead of the next rotation. This is the most strategically rich mode.

## Scale Up

The arena grows as the match heats up.

* Starts at 3x3. Win condition stays "3 in a row" at all sizes.
* After **round 4**, the board expands to 4x4. Existing pieces are preserved in the top-left quadrant; new cells are empty.
* After **round 9**, the board expands to 5x5. Same expansion logic.
* The on-chain `board` field is a flat `[CellState; 25]`. The active dimension lives in `board_size`.
* The win-detection algorithm slides a 3-cell window over every row, column, and diagonal — works identically at every size.

## Blitz (Phase 2)

For players who want it decisive and fast.

* Each turn has a hard **5-minute (300 second) on-chain timeout** instead of 24 hours.
* If a reveal is submitted after the 300-second window, the program treats it as a wrong answer: the hash is cleared, no piece is placed, the turn passes.
* `timeout_turn` becomes the cleanup tool when an opponent rage-quits or zones out.

## Ultimate TTT (Stretch)

A 9x9 meta-grid where each cell of the outer grid is itself a 3x3 TTT board. Win the inner game to claim the outer cell. Win three outer cells in a row to win the match.

This requires a redesigned `GameAccount` layout and a separate set of instructions. Tracked under V2.0 on the [Roadmap](/mindduel-docs/resources/roadmap.md).

## How a mode is selected

The mode is chosen at match creation and stored on-chain in `GameAccount.mode`. The contract enforces the mode-specific logic in `reveal_answer` — the frontend cannot lie about which mode is active.

Frontend mode IDs map to on-chain enums as follows:

| Frontend ID | On-chain enum             |
| ----------- | ------------------------- |
| `classic`   | `GameMode::Classic`       |
| `shifting`  | `GameMode::ShiftingBoard` |
| `scaleup`   | `GameMode::ScaleUp`       |
| `blitz`     | `GameMode::Blitz`         |
| `vs-ai`     | (off-chain only)          |


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://mindduel.gitbook.io/mindduel-docs/features/game-modes.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.